Table 2 Comparison of the probabilities for the magnitudes of the largest expected aftershocks to be larger than m = Mms − 1.0 for several prominent past aftershock sequences

From: Forecasting the magnitude of the largest expected earthquake

Date

Name

M ms

m 0

ETAS (Gamma)

OU

OUMCMC

EVD

m as

2018/11/30

Anchorage

7.1

2.5

0.10

0.08

0.12

0.17

4.8

2016/11/13

Kaikoura

7.8

3.7

0.10

0.40

0.49

0.52

5.7

2016/04/16

Kumamoto

7.3

3.3

0.24

0.15

0.19

0.26

5.8

2011/03/11

Tohoku

9.0

5.3

0.27

0.15

0.19

0.26

6.7

2010/04/04

El Mayor

7.2

3.3

0.24

0.15

0.23

0.22

5.3

2008/06/14

Iwate

7.2

3.1

0.06

0.06

0.08

0.11

5.3

2007/03/25

Noto

6.9

3.1

0.15

0.13

0.17

0.23

4.9

2005/03/20

Fukuoka

7.0

3.1

0.02

0.01

0.02

0.03

5.4

2003/09/26

Tokachi-oki

8.0

3.3

0.27

0.23

0.28

0.40

6.5

2002/11/03

Denali

7.9

3.0

0.08

0.24

0.38

0.47

5.5

  1. The Bayesian predictive distribution PB(mex > m|S, ΔT) was computed using: ETAS—the ETAS model with the Gamma prior for the model parameters; OU—the method developed in ref. 4 with the Omori–Utsu law and a flat prior; OUMCMC—the MCMC sampling with the Omori–Utsu law and the Gamma prior. These are compared with the probabilities computed using the Gumbel distribution as the EVD for the non-homogeneous Poisson process with the parameters of the Omori–Utsu law estimated using the maximum-likelihood method. The training time interval [T0, Te] = [0.0, 2.0] days and the forecasting time interval ΔT = 10 days were used, where T0 = 0 corresponds to the time of the occurrence of the mainshocks. The last column gives the magnitudes of the actual largest aftershock that occurred during the forecasting time interval ΔT = 10 days
Back to article page